cogpy.decomposition.pca

ERP-PCA decomposition of neural signals.

Provides varimax-rotated PCA for spatio-spectral decomposition of ECoG data. The erpPCA class follows the scikit-learn estimator API (fit / transform).

Functions

compute_similarity_matrix(ss1, ss2[, ...])	Pairwise similarity matrix between factors of two SpatSpec instances.
erpPCA2factors(erp, mtx_sws)	Convert fitted erpPCA + spectrogram into (ldx, scx) DataArrays.
erpPCA_ldx(LR, ld_coords, ld_shape)	Reshape rotated loadings into an xr.DataArray (factor, h, w, freq).
erpPCA_scx(FSr, times)	Wrap factor scores into an xr.DataArray (time, factor).
erppca(X[, nFac, maxIt, Tol, IfVerbose, ...])	Run ERP-PCA: covariance → eigen → varimax → sort.
get_invproj(ldx)	Compute inverse projection from loading DataArray (factors, h, w, freq).
get_similarity(ss_series, freq_threshold)	Build an (nrec, nrec) array of (nfac, nfac) similarity matrices.
kaiser_normalization(X)
project_to_loadings(X, inv_proj)	Project data onto pre-computed inverse projection.
pseudo_inverse(LR, cov)
pseudo_inverse_scaled(LR, cov)
redirect_loadings(L)
similarity_metric(ss1, ss2, fac1, fac2[, ...])	Correlation-based similarity between two SpatSpec factors.
simplicity_criterion(Y)	Simplicity criterion for varimax columns.
sort_by_eigv(L, ev)
spatspec2erpPCA(ss)	Create an erpPCA estimator from a SpatSpecDecomposition.
varimax_iter(Yo, Y)
varimax_rotation(X[, maxit, tol, norm, ...])	Varimax rotation (4M algorithm).

Classes

erpPCA([nfac, max_it, tol, verbose, ...])

Varimax-rotated PCA estimator for ECoG spatio-spectral data.